ANIMAL BEHAVIOUR, 2000, 59, 577–586 doi:10.1006/anbe.1999.1326, available online at http://www.idealibrary.com on The diving behaviour of green turtles at Ascension Island GRAEME C. HAYS*, COLIN R. ADAMS†, ANNETTE C. BRODERICK*, BRENDAN J. GODLEY*, DAVID J. LUCAS*, JULIAN D. METCALFE‡ & ANDREA A. PRIOR* *School of Biological Sciences, University of Wales Swansea †University of Glasgow Veterinary School ‡The Centre for Environment, Fisheries and Aquaculture Science, Lowestoft Laboratory (Received 14 July 1999; initial acceptance 29 September 1999; final acceptance 6 November 1999; MS. number: 6286) For six green turtles, Chelonia mydas, that had nested on Ascension Island in the South Atlantic, we used time–depth recorders to examine their diving behaviour during the subsequent internesting interval (10–12 days). All the turtles performed dives where they remained at a fixed depth for a long period, surfaced briefly and then dived to the same depth again. It is generally believed these dive profiles are caused by the turtles resting on the sea bed. The maximum depth that turtles routinely reached on these resting dives was between 18 and 20 m, with resting dives deeper than 20 m being extremely rare. Resting dive duration increased significantly with deeper dives. From this relationship, and assuming that turtles with fully inflated lungs at the surface need to dive to 19 m to achieve negative buoyancy, we estimated for two turtles that the oxygen consumption during resting dives was 0.016 and 0.020 litres O 2 /kg per h, respectively. This is similar to the value predicted from the allometric scaling relationship for the minimal oxygen consumption of turtles. We calculated that the energy conserved by resting during the internesting period may appreciably increase the reproductive output of females.  2000 The Association for the Study of Animal Behaviour A fundamental component of many ecological studies is how animals allocate time to different behaviours. This area of science is particularly well developed for terrestrial species since they can often be readily observed. In contrast, marine animals present a more intractable prob- lem since they may remain submerged, and hence out of view, for long periods. This dichotomy between terrestrial and marine studies is reflected in our knowledge of sea turtles in their natural habitats. On the one hand, obser- vations on nesting beaches have been relatively straight- forward to make, leading to many studies that have examined aspects such as body size, clutch and egg size, nesting behaviour, hatchling success and individual nest- ing periodicity. On the other hand, studies of sea turtles while they are at sea have traditionally been more diffi- cult to make and so have been limited to sporadic obser- vations of their behaviour and fragmentary evidence of long-distance movements from mark–recapture studies (for example see Dodd 1988 for a review). This imbalance in information is, however, starting to be redressed with the advent of small, reliable data-logging and tracking devices. For example, satellite tracking has identified the routes that turtles follow during their migrations between nesting and feeding areas (e.g. Luschi et al. 1998) while implanted tempera- ture loggers have been used to examine aspects of their thermoregulatory capacity (Sato et al. 1994). Similarly, the advent of reliable time–depth recorders has begun to allow the diving behaviour of free-ranging turtles to be documented (Eckert et al. 1989). These recorders have, for example, been used to distinguish periods of inactivity, where turtles remain resting on the sea floor for long periods, from bouts of activity where the depth of the animal varies constantly (van Dam & Diez 1996). It has been shown experimentally that, for the loggerhead turtle, Caretta caretta, individuals control their buoyancy by varying the volume of air inspired immediately prior to submergence (Milsom 1975). Consequently, Milsom (1975) suggested that when the turtle embarks on a deep dive, a greater volume of air is inspired so that neutral buoyancy is attained at a greater depth. As a corollary to this hypothesis, it would be predicted that if more air is inspired before deeper dives, then the available oxygen stores will be greater and hence dive duration will increase. Data from time–depth recorders have, indeed, shown that for loggerhead turtles dive duration increases Correspondence: G. Hays, School of Biological Sciences, University of Wales, Swansea SA2 8PP, U.K. (email: g.hays@swan.ac.uk). C. R. Adams is at the University of Glasgow Veterinary School, Bearsden Road, Glasgow G61 1QH, U.K. J. D. Metcalfe is at The Centre for Environment, Fisheries and Aquaculture Science, Lowestoft Laboratory, Lowestoft NR33 0HT, U.K. 0003–3472/00/030577+10 $35.00/0  2000 The Association for the Study of Animal Behaviour 577